I'm trying to run prefect from docker, but can't get the token through. Somehow I'm getting:

prefect.utilities.exceptions.ClientError: Malformed response received from Cloud - please ensure that you have an API token properly configured.

Anyone?

Hi, I was wondering if prefect always keep its task results values in memory during whole flow run or if gets removed from memory when enabling writing to a LocalResults target. As an example, what happens in the following task?

jsonresult = task(
        get_json_fun,
        result=LocalResult(serializer=JSONSerializer()),
        target='/path/result.json',
        checkpoint=True,
    )(p)

Is jsonresult still in memory next to being written to result.json or not? I'm worried that my machine will get out of memory when my flow get's bigger with many task results floating around.

New to GraphQL and trying to use the Prefect Interactive API; this is my query but I am getting an error, can anyone help? I have a feeling there is something super simple I am missing: query:

query {
  flow_run_state {
    flow_run {
      id,
      parameters,
      created,
      flow(where:{name:"my-flow-name"}) {
        name,
        id
      },
    }
  }
}

Error:

{
  "graphQLErrors": [
    {
      "message": "Unknown argument \"where\" on field \"flow\" of type \"flow_run\".",
      "extensions": {
        "code": "GRAPHQL_VALIDATION_FAILED"
      }
    }
  ],
  "networkError": null,
  "message": "GraphQL error: Unknown argument \"where\" on field \"flow\" of type \"flow_run\"."
}

Hi! I’m trying to run prefect server on virtual centos machine with custom .prefect/config.toml

[server]

host = "ip"
port = "8081"
url = "https://${server.host}:${server.port}"

[ui]

host = "<http://ip>"
port = "8081"
host_port = "8081"
endpoint = "${ui.host}:${ui.port}"

And after checking prefect.config in python, I see my custom properties:

<Box: {'debug': False, 'home_dir': '/home/o.ilinsky/.prefect', 'backend': 'server', 'server': {'host': 'ip', 'port': 8081, 'host_port': 4200, 'endpoint': 'ip:8081', 'database': {'host': 'localhost', 'port': 5432, 'host_port': 5432, 'name': 'prefect_server', 'username': 'prefect', 'password': 'test-password', 'connection_url': '<postgresql://prefect:test-password@localhost:5432/prefect_server>', 'volume_path': '/home/o.ilinsky/.prefect/pg_data'}, 'graphql': {'host': '0.0.0.0', 'port': 4201, 'host_port': 4201, 'debug': False, 'path': '/graphql/'}, 'hasura': {'host': 'localhost', 'port': 3000, 'host_port': 3000, 'admin_secret': '', 'claims_namespace': 'hasura-claims', 'graphql_url': '<http://localhost:3000/v1alpha1/graphql>', 'ws_url': '<ws://localhost:3000/v1alpha1/graphql>', 'execute_retry_seconds': 10}, 'ui': {'host': '<http://localhost>', 'port': 8080, 'host_port': 8080, 'endpoint': '<http://localhost:8080>', 'graphql_url': '<http://localhost:4200/graphql>'}, 'telemetry': {'enabled': True}, 'url': '<https://ip:8081>'}, 'cloud': {'api': 'ip:8081', 'endpoint': '<https://api.prefect.io>', 'graphql': 'ip:8081/graphql', 'use_local_secrets': True, 'heartbeat_interval': 30.0, 'check_cancellation_interval': 15.0, 'diagnostics': False, 'logging_heartbeat': 5, 'queue_interval': 30.0, 'agent': {'name': 'agent', 'labels': [], 'level': 'INFO', 'auth_token': '', 'agent_address': '', 'resource_manager': {'loop_interval': 60}}}, 'logging': {'level': 'INFO', 'format': '[%(asctime)s] %(levelname)s - %(name)s | %(message)s', 'log_attributes': [], 'datefmt': '%Y-%m-%d %H:%M:%S', 'log_to_cloud': False, 'extra_loggers': []}, 'flows': {'eager_edge_validation': False, 'run_on_schedule': True, 'checkpointing': False, 'defaults': {'storage': {'add_default_labels': True, 'default_class': 'prefect.environments.storage.Local'}}}, 'tasks': {'defaults': {'max_retries': 0, 'retry_delay': None, 'timeout': None}}, 'engine': {'executor': {'default_class': 'prefect.engine.executors.LocalExecutor', 'dask': {'address': '', 'cluster_class': 'distributed.deploy.local.LocalCluster'}}, 'flow_runner': {'default_class': 'prefect.engine.flow_runner.FlowRunner'}, 'task_runner': {'default_class': 'prefect.engine.task_runner.TaskRunner'}}, 'ui': {'host': '<http://ip>', 'port': 8081, 'host_port': 8081, 'endpoint': '<http://ip:8081>', 'graphql_url': '<http://ip:4200>'}}>

But server starts with localhost:8080 option:

Visit <http://localhost:8080> to get started

And nothing on 8081 port. I also tried to set env

export PREFECT__USER_CONFIG_PATH=/home/o.ilinsky/.prefect/config.toml

But it’s does’t work too. Do you know what I do wrong?) P.S. i’m new in prefect, so it may by stupid error :) ``````

hello! Has anyone found an s3 task example using prefect cloud ? tad confusing on how to authenticate + download a file. Do we use S3ResultHandler + cloud secret >> then pass this to S3Download task ?

Hi, are there any examples of testing out the config functionality from unit tests? I tried the below but config.environment yields a BoxKeyError ("Config object has no attribute environment")

def test_configuration_works_as_expected(monkeypatch, tmp_path):
    config_file = tmp_path / "config.toml"
    config_text = """
environment = "prod"
user = "${environments.${environment}.user}"

[environments]

    [environments.dev]
        user = "test"

    [environments.prod]
        user = "admin"
"""
    with config_file.open("w") as cf:
        cf.write(config_text)

    monkeypatch.setenv("PREFECT__USER_CONFIG_PATH", str(tmp_path))
    assert config.environment == "prod"
    assert config.user == "admin"

Anyone ran into weird behaviour with Task Concurrency? When the limit is 4, it will use all of these available slots (expected behaviour), but as soon as we set 6 or 8 it only allows 2 tasks to run at a time (not 100% of the potential slot capacity). The difference in total flow runtime (concurrency = 4 vs 8 ) also follows the pattern, of course

Hello everyone, I have three schedulers that trigger the same flow. Each scheduler is passing different

parameters_default

arguments. The problem is that the third scheduler never triggers a flow run and his

parameters_default

arguments are never passed to any flow run. Is this some kind of conflict between the second and the third scheduler (same time)? I'm pretty sure that I have misunderstand how the scheduling is working. Thanks in advance

Hey everyone, I'd like to start a

FlowRunTask

based on the result of a previous task. To be precise: Let's say I have a task and that task needs to fetch data from an API but hits the API request limit, now I want the task to return the data that it was able to fetch but also some sort of flag that shows that the task couldn't fetch all data because the api limit was hit, e.g. in a dict

{"data": data, "all_data_fetched": False}

. Now if

all_data_fetched

is false I'd like to start a new flow that continues fetching the data one hour later since that's the time it takes until the api limit is reset. I want to create a new flow since we're running on fargate and I don't want to let the task wait for one hour and pay all the time. And at the same time I want the rest of my flow to continue processing that data that the task was able to fetch from the API. I hope a) it's somewhat understandable what I want to achieve, b) it's possible and c) there is someone out there who can give me a hint on how to implement it. My biggest issue is that I don't know how to access the result from a task within a flow to check if I need to trigger the

FlowRunTask

or not. AFAIK I cannot run

FlowRunTask

from within a task, right? TIA 🙂

Hi y’all. Can folks tell me whether it’s necessary to know/learn all the ins and outs of Airflow before using Prefect? I’m fairly new to DAGs — I’ve only ever dabbled with Airflow in short video tutorials. Ideally, I would skip it entirely and just use Prefect at my current job, if the learning curve isn’t too steep and/or expects that you’ll understand all the Airflow-related things before using it. Thanks!

Hi Everyone, In prefect cloud - I’m curious if there is a way to either restrict what input parameter values can be entered or see what options are available for a flow? I was just looking at your arithmetic example where you’re using a switch to set available math operations to pick from. I guess in prefect cloud, you would need to know ahead of time what options you can use? https://docs.prefect.io/core/advanced_tutorials/calculator.html

Is there any sort of built-in support for task arguments that can be either a

Result

instance or a raw value? The functionality I’m looking for is basically this:

@task
def foo(bar):
    if isinstance(bar, Result):
        bar = bar.read(bar.location).value
    return do_something(bar)

Clearly the above isn’t hard to add, but it also seems like the kind of thing where (a) there might be hidden gotchas and (b) it seems common enough that

prefect

might have something built in

Hi Everyone, hoping I am overlooking something simple. My drive filled up in my test environment, due to the results folder caching everything. I am running Prefect Core Server (v0.13.5) and I currently have checkpointing disabled. I read the docs here https://docs.prefect.io/core/advanced_tutorials/using-results.html and I believe I have everything configured properly. How can I a) make sure checkpointing is really disabled and b) change the global location of the results directory?

Hi, looking into Prefect now and I’m falling in love. I have a question regarding subscribing to Queues like GCP-PubSub, RabbitMQ, etc. Subscribing and reacting when something is put on the queue is very convenient for a lot of usecases. Does Prefect have some sort of subscriber functionality?

hi guys for google container registry authentication i got it to work with docker agent which can pull images for flows using docker storage but the flow containers themselves seem to fail to pull images how is the docker agent picking up the credentials? from my experimentation it seems to look at

~/.docker/config.json

but does this mean for flows i have to pack the credentials in the image, instead of being able to mount a volume when the container is run? i have looked at API documentation and couldn’t find any options for specifying volumes for flow containers

Hi , I have a question about combining parametrized flow runs into one flow. Assume this pseudocode:

with Flow('flow-a') as flowA:
    paramsFlowA = Parameter('flow-a-param', default=...)
   < Does something here > 
with Flow('flow-b') as flowB:
   paramsFlowB = Parameter('flow-b-param', default=...) 
  < Does something here> 
with Ffow('combined-flow') as combined_flow: 
   flow_a = FlowRunTask(flow_name='flow-a', project_name='test') 
   flow_b = FlowRunTask(flow_name='flow-b', project_name='test) 

   result = flow_b(upstream_tasks[flow_a])

When I deploy the combined_flow to the server, I can't set the parameters for FlowA and FlowB. Is it possible to do so and how? Thanks!

hi guys with a prefect function task i can return a result simply with

return

like this:

@task
def start_job():
   return job_id

i was wondering if it was possible to do the same for a container task created with

CreateContainer

in my case what i am trying to do is to return the job id of the job started by that container, so that another task can use it to stop the job

Hi all, It may seems a stupid question, but I can’t find a way to print/log the name and the value of a parameter inside a Flow. How is it supposed to be done? Any help and directions is much appreciated 🙂

hi guys is it possible to have a task have both data and non-data dependencies?

upstream_tasks=[upstream_task]

only seems to work when there are no data-dependencies?

hi everyone, how can I set AWS_CREDENTIALS from the Secrets GUI?

I don't see a way to set a key/value pair there

seeing some typos in the docs, parameter is databricks_conn_secret but examples say databricks_conn_string

Hello all! I have kind of a generic question. As anyone tried to use sqlalchemy mappers to store Prefect Tasks in a Postgres database or similar? Is this possible with Prefect Task classes? Something similar to how Airflow creates its tables

Hi there,

I was searching for a paradigm where a task will create small results and another task will take that data and start instantly. So something like a flow (A -> B) where A, B are tasks and A will

yield

result for B. B will start when it gets first

yield

from A

I'm trying to figure out how to run some Tasks in parallel, like something like this in the second stage. There is no data passed between the stages, the tasks in the previous stage just have to complete with Success I know how to run them sequentailly like task_2(task_1()) but unclear if I should be using .map?

Hi everyone 👋, Is it possible to create a task which is triggered based on some external event (e.g. email received, or website updated specific data point, etc)? I wonder how would one go about creating such a flow?:) 🤔